AI security startup Lasso Security has introduced LEAP, a CPU-only guardrail for AI applications that eliminates the need for expensive GPU acceleration. The company also announced a $30 million funding round to expand engineering and sales operations, targeting regulated and federal markets.
- LEAP runs AI guardrails on standard CPUs with milli-second latency
- New $30M funding boosts engineering and market expansion efforts
- Platform deployed by major firms including BMW and US Homeland Security
Market signal
Lasso Security’s launch of LEAP reflects a growing demand in the enterprise technology market for scalable, cost-effective AI security solutions that do not require expensive GPU infrastructure. Traditional AI guardrails have relied on secondary large language models running on accelerated hardware to vet AI outputs, driving up costs and limiting thorough traffic inspection. LEAP’s CPU-only architecture largely removes this barrier, enabling inspection at thousands of times the throughput of prior solutions.
The $30 million Series A funding, led by ClearSky Advisors with participation from previous backers Entrée Capital, iAngels, and others, validates investor confidence in the viability and market readiness of CPU-driven AI guardrails. This positions Lasso Security not only as a technology innovator but also as a growing competitor in the high-stakes AI security segment serving heavily regulated industries worldwide.
Operator impact
For operators deploying AI systems at scale—especially in sectors like financial services, healthcare, insurance, and government—the cost and complexity of existing AI guardrails has limited comprehensive traffic monitoring. Lasso’s LEAP significantly reduces operational expenses by processing AI risk checks on CPUs and bypassing the need for costly GPU resources for most AI traffic. The platform’s rapid verdict response (under five milliseconds) enables real-time AI security enforcement without throughput bottlenecks.
The dual-engine platform model, including the RAPID self-hosted engine for nuanced, policy-based judgments, provides a layered defense approach. Operators benefit from routing higher-risk queries to more sophisticated AI models only when necessary, optimizing resource allocation. The real-world application in environments like the US Department of Homeland Security and BMW demonstrates LEAP’s ability to secure billions of AI prompts monthly, supporting enhanced risk management without prohibitive investment.
What to watch next
Lasso Security’s expansion plans into federal, regulated-industry accounts, and broader North American and European markets will be critical to follow, as regulatory scrutiny and operational AI risk management continue to intensify globally. Their ability to scale engineering and sales teams effectively from Tel Aviv and New York will influence adoption rates and competitive positioning against legacy AI security providers reliant on GPU-heavy architectures.
Additionally, the integration of continuous offensive testing into the product development cycle—automated red teaming and autonomous attack simulations—points toward a trend of proactive AI security built on adversarial insights. Observers should monitor how this approach advances AI guardrail sophistication and whether competitors follow suit with similar disruptive technologies that promise high throughput, low cost, and tighter security compliance in enterprise AI deployments.